SU-F-J-54: Towards Real-Time Volumetric Imaging Using the Treatment Beam and KV Beam

Authors


Abstract

Purpose:

Existing real-time imaging uses dual (orthogonal) kV beam fluoroscopies and may result in significant amount of extra radiation to patients, especially for prolonged treatment cases. In addition, kV projections only provide 2D information, which is insufficient for in vivo dose reconstruction. We propose real-time volumetric imaging using prior knowledge of pre-treatment 4D images and real-time 2D transit data of treatment beam and kV beam.

Methods:

The pre-treatment multi-snapshot volumetric images are used to simulate 2D projections of both the treatment beam and kV beam, respectively, for each treatment field defined by the control point. During radiation delivery, the transit signals acquired by the electronic portal image device (EPID) are processed for every projection and compared with pre-calculation by cross-correlation for phase matching and thus 3D snapshot identification or real-time volumetric imaging. The data processing involves taking logarithmic ratios of EPID signals with respect to the air scan to reduce modeling uncertainties in head scatter fluence and EPID response. Simulated 2D projections are also used to pre-calculate confidence levels in phase matching. Treatment beam projections that have a low confidence level either in pre-calculation or real-time acquisition will trigger kV beams so that complementary information can be exploited. In case both the treatment beam and kV beam return low confidence in phase matching, a predicted phase based on linear regression will be generated.

Results:

Simulation studies indicated treatment beams provide sufficient confidence in phase matching for most cases. At times of low confidence from treatment beams, kV imaging provides sufficient confidence in phase matching due to its complementary configuration.

Conclusion:

The proposed real-time volumetric imaging utilizes the treatment beam and triggers kV beams for complementary information when the treatment beam along does not provide sufficient confidence for phase matching. This strategy minimizes the use of extra radiation to patients.

This project is partially supported by a Varian MRA grant.

Ancillary